The rapid ascent of antibody-drug conjugates (ADCs) has transformed the oncology landscape, shifting the paradigm from traditional, systemic chemotherapy to a model of "guided missiles" for cancer treatment. With roughly 15 FDA-approved products currently in circulation and a market projected to reach $57 billion by 2032, the pressure to optimize these therapies is immense. Yet, a persistent clinical enigma remains: why do tumors that test positive for a drug’s target still frequently resist treatment?
This critical question has sparked a strategic collaboration between pharmaceutical giant Gilead Sciences and Nucleai, a leader in AI-driven spatial biology. By pivoting from conventional pathology to advanced computational analysis of clinical trial data, the two organizations are attempting to bridge the gap between what a pathologist sees under a microscope and what a drug actually encounters within the chaotic, hostile environment of a tumor.
The "Target-Positive" Paradox
At its inception, the ADC promise was elegantly simple: use an antibody as a GPS system to ferry a potent cytotoxic payload directly to the tumor cell, sparing healthy tissue. However, Dr. Ken Bloom, head of pathology at Nucleai, notes that clinical reality is far more complex.
"Initially, what we were told was that ADCs are basically targeted chemotherapy," Bloom explains. "The promise is getting all the benefits of chemotherapy without the toxicity. But in practice, it’s far trickier than that."
The "tricky" reality involves several failure modes. A tumor may appear "target-positive" on a standard immunohistochemistry (IHC) slide, but the target itself may be structurally altered, glycosylated, or physically inaccessible to the bulky ADC molecule. Furthermore, the tumor microenvironment (TME) is a minefield of extracellular proteases capable of prematurely cleaving ADC linkers, while cancer cells themselves have evolved sophisticated efflux pumps to expel the cytotoxic payload once it enters the cell.
Chronology of a Collaborative Effort
The partnership between Gilead and Nucleai did not emerge overnight; it is the product of a multi-year maturation process in translational medicine.
- Initial Phase (2022–2023): The relationship began with a focus on preclinical multiplex immunofluorescence (mIF). Nucleai provided the technical infrastructure to map the spatial distribution of biomarkers in early-stage research, moving beyond simple expression counts to examine the cellular architecture of tumors.
- Expansion (2024): Recognizing the limitations of standard biomarker assays, the collaboration deepened. Gilead began integrating Nucleai’s platform into its clinical trial pipeline, moving from the lab bench to the analysis of retrospective patient samples.
- The August 2026 Disclosure: On August 11, 2026, the companies publicly confirmed that Nucleai had completed extensive analysis of H&E (hematoxylin and eosin) and IHC whole-slide images from multiple Gilead oncology clinical studies. This phase aimed to link complex tissue features directly with clinical outcomes, such as progression-free survival and objective response rates.
The Inside-Outside Epitope Problem: A Technical Barrier
The core of the issue lies in a "hidden secret" of pathology: the bias of current detection methods. Most antibodies used in standard diagnostic IHC are designed to bind to the intracellular domain of a protein because those regions are better preserved during tissue processing.
However, ADCs must bind to the extracellular epitope of the target receptor to initiate internalization. A classic example is HER2. The industry-standard Ventana 4B5 IHC antibody detects the intracellular domain, meaning it can flag cells that express HER2 fragments (p95HER2) that lack the very extracellular "docking station" required for drugs like trastuzumab.
"There’s a reason that we look at the internal side as pathologists—we try to stay inside the membrane," says Bloom. "Tissue processing tends to preserve the intracellular portions better. Yet it’s the outside that you should really care about."
By relying on assays that quantify the wrong side of the membrane, clinicians may be misclassifying patients as "positive" for a drug that has no physical way to attach to their cancer cells.
Computational Pathology: Mapping the "Neighborhood"
To overcome these limitations, Nucleai is applying AI-driven spatial intelligence. Unlike traditional digital pathology, which simply digitizes a slide for remote viewing, computational pathology extracts multidimensional data from the tissue.
Spatial Relationships and Cellular Density
The platform examines the "neighborhood" of a tumor cell. It asks not just if the target is present, but:
- What is the spatial distribution of the target within the tumor?
- How dense is the target expression in specific tumor regions versus the surrounding stroma?
- What is the proximity of the target-positive cells to immune cells or vascular structures?
By quantifying these relationships, the researchers can identify "spatial biomarkers"—patterns of tumor architecture that are significantly more predictive of treatment success than the binary "positive/negative" scoring systems of the past.
Official Perspectives: The Synergy of Human and AI
For Gilead, this collaboration is a logical extension of its commitment to precision oncology. According to Meghna Das Thakur, senior director of oncology biomarkers at Gilead, the collaboration aims to enhance the efficiency of biomarker discovery. "Combining [Nucleai’s] specialized expertise in AI-driven spatial biology and tissue analytics with Gilead’s scientific and clinical expertise allows us to generate insights more efficiently," she stated.
Dr. Bloom emphasizes that the goal is not to replace the pathologist, but to augment their capabilities. He frequently cites the mantra, "A fool with a tool is still a fool."
"AI by itself isn’t going to do it, but a well-trained physician with AI is going to be light-years ahead of one without," Bloom notes. He argues that computational tools reduce the variability inherent in human fatigue and subjective grading, effectively "raising the floor" of diagnostic reliability. By color-coding tumor versus non-tumor tissue and providing spatial heatmaps, the AI creates a common language that oncologists, surgeons, and pathologists can all interpret with high fidelity.
Implications for the Future of Oncology
The findings from this collaboration are expected to emerge in peer-reviewed literature by late 2026 or early 2027. Early results already suggest that several candidate biomarkers have surfaced—features that, while visible on standard H&E or IHC slides, were previously ignored because they were too complex for the human eye to quantify consistently.
The "Wave" Effect
The industry is currently watching Gilead and Nucleai closely. As Bloom suggests, there is no clearly defined, industry-wide "pathway" yet for integrating computational pathology into the regulatory submission process for new drugs. However, the first company to successfully demonstrate that AI-derived spatial biomarkers correlate with clinical trial endpoints will likely set a new gold standard.
"What you’re going to see, as soon as the first one hits, is a wave that follows," Bloom predicts.
Precision Medicine Beyond the Binary
The implications of this work extend far beyond Gilead’s current ADC pipeline. If researchers can prove that spatial context and epitope accessibility are more critical than raw expression levels, it could necessitate a wholesale redesign of how we screen for cancer therapies.
For the patient, this represents the potential for fewer failed treatments, better allocation of high-cost precision drugs, and a more nuanced understanding of why specific therapies work for some individuals while failing others. By evolving from "looking at the slide" to "analyzing the ecosystem," the partnership between Gilead and Nucleai is turning the opaque landscape of the tumor microenvironment into a transparent, actionable map for the next generation of cancer therapy.
About the Author
